Batch correction (mass spectrometry data)

Question

[mkhadka10]

Hi,

I used my raw data for batch correction using Combat package. It worked well before normalization. However after normalization of batch corrected data, the correction is removed and my data showed clustering based on batch processed.

Could anyone please tell me if I have to normalize data before batch correction? Is it correct to use batch corrected data for normalization and further downstream processing?

Let me know if my question is not correct.

Thank you

[mkhadka10]

Thank you for your response.

[GenoMax]

Please use ADD REPLY/ADD COMMENT when responding to existing posts. It helps keeps threads logically organized.

It would be useful to add a sentence or two as to whether @ddiez's response was helpful, in the right direction or completely off-target. You should also up-vote and/or accept (use the checkmark against the post) the answer, which provides acknowledgement for the person who answers your question.

Answer 1

[ddiez]

There are several alternative ways (or workflows) to deal with batch correction. I am assuming that you are using the bioconductor package sva. There are several examples in the vignette explaining how to analyze the data, in particular sections 7 and 8. Personally, I preprocess/normalize the data first, then include the batch information as a variable in the model (typically linear model with limma package).